Water conservancy project construction intelligent inspection method and system based on unmanned aerial vehicle

Abstract

The invention relates to the technical field of water conservancy project construction, and discloses a water conservancy project construction intelligent inspection method and system based on an unmanned aerial vehicle, and the method comprises a route planning step, an inspection preparation step, a data collection step, a data recognition step, a data uploading step and a closed-loop management step. The system corresponds to the method. According to the invention, through full-automatic inspection of the unmanned aerial vehicle, full-coverage imaging and special detection are carried out on construction key parts and risk areas, so that no-dead-corner inspection of the construction areas is realized; defects are automatically recognized based on an AI recognition system, positions are marked and graded, an inspection log is generated, and the defect recognition efficiency and accuracy are improved; secondly, the ground terminal uploads the data to a supervision platform, the supervision platform carries out filing according to construction nodes and generates a closed-loop tracking list, and an information island is broken; therefore, the supervision accuracy, timeliness and traceability are improved, the visual, quantifiable and traceable supervision requirements of the construction period of the water conservancy project are met, and the construction quality and safety are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydraulic engineering construction, in particular to a method and system for intelligent inspection of hydraulic engineering construction based on unmanned aerial vehicles. BACKGROUND

[0002] With the continuous expansion of the scale of hydraulic engineering construction, the construction environment of major infrastructure projects such as dams and long-distance water pipelines is becoming more and more complex, and the quality and safety supervision during the construction period has become a key link to ensure the success of the project construction. At present, the inspection during the construction period of water conservancy projects mainly relies on traditional manual inspection methods, which has significant limitations:

[0003] Manual inspection is limited by terrain (such as deep foundation pits, high slopes, and water areas), making it difficult to achieve full-space dead-angle coverage, and when conducting inspections in high-risk operation areas (such as high-altitude hoisting areas and deep foundation pit slopes), it is easy to cause safety accidents to construction personnel;

[0004] The inspection results are highly dependent on personnel experience, and the identification accuracy of subtle issues such as structural deviation, weld defects, and leakage traces is low, making it easy to miss or misjudge, leading to small problems evolving into major failures;

[0005] Inspection data is mainly recorded manually and in scattered images, lacking standardized archiving and correlation positioning, and the data transmission path is long, from problem discovery to feedback and rectification, which takes several hours to several days, making it difficult to meet the needs of timely disposal during the construction period;

[0006] Key parts during the construction period (such as dam pouring layers and pipeline interfaces) change dynamically with progress, and traditional inspection routes are fixed, making it impossible to intensify supervision for key processes (such as concrete curing and impermeable membrane laying), and making it easy to have blind spots in supervision.

[0007] In the prior art, unmanned aerial vehicles have been applied to water conservancy project inspection, but most focus on the operation and maintenance stage after completion (such as structure aging detection), however, on the one hand, there is no linkage mechanism between construction progress and route adjustment, and the route coverage range cannot match the dynamic construction nodes; on the other hand, there is a lack of customized AI identification models during the construction period, and general models cannot accurately identify defects specific to the construction period (such as formwork deviation and welding misalignment), and inspection data is not deeply integrated with enterprise supervision platforms, highlighting the problem of information silos.

[0008] In summary, the existing inspection technology cannot meet the needs of visual, quantifiable, traceable, and high-response supervision during the construction period of water conservancy projects, and there is an urgent need for an inspection solution that can adapt to construction dynamics, intelligently identify defects, and achieve data closure. SUMMARY

[0009] The present application aims to provide a method and system for intelligent inspection of hydraulic engineering construction based on unmanned aerial vehicles to solve the technical problems raised in the background.

[0010] To achieve the above object, the application discloses the following technical solutions:

[0011] In a first aspect, the application discloses a water conservancy project construction intelligent inspection method based on a UAV, which comprises the following steps:

[0012] A flight path planning step: based on the imported construction-related drawings, the key construction positions are analyzed, the risk areas are determined based on the construction-related drawing analysis, construction scheme risk marking or historical hidden danger records, the inspection flight path is generated for the key construction positions and the risk areas, and the coverage range of the inspection flight path is dynamically adjusted according to the construction progress, and the generated inspection flight path is uploaded to the flight control terminal of the UAV after being audited and confirmed;

[0013] An inspection preparation step: checking the working state of the power supply equipment, flight parts, sensing parts and gimbal of the UAV, and calibrating the positioning module of the UAV;

[0014] A data collection step: the UAV starts the full-automatic inspection mode and follows the inspection flight path to perform full-coverage imaging collection on the key construction positions through the imaging equipment carried thereon, and performs special detection on the risk areas through the detection equipment carried thereon; after the flight task is suspended, the ground terminal performs preliminary screening on the collected image data, controls the UAV to take supplementary pictures for the unqualified areas, and after the supplementary pictures are taken, the UAV continues to perform the remaining inspection task or ends the flight task;

[0015] A data recognition step: after the flight task is completed, the qualified images are imported into an AI recognition system, the AI recognition system identifies defects based on a preset defect library, marks the defect positions, classifies the defects according to the defect severity, and generates an inspection log based on the recognition result;

[0016] A data uploading step: the ground terminal integrates the image data, the inspection log and the associated positioning data to generate an inspection data package, and uploads the inspection data package to a supervision platform after being encrypted;

[0017] A closed-loop management step: the supervision platform archives the inspection data package according to the construction node, pushes the serious defects to the construction unit and the supervision party responsible person, records the response time and processing progress of the defects, and generates a defect disposal closed-loop tracking sheet; wherein the electronic log generated each time is retained for no less than a specified period of time.

[0018] Preferably, the determination of the risk area comprises at least one of the following:

[0019] Based on the construction-related drawing analysis: the geological condition data, structure stress weak points and key process nodes in the construction-related drawings are extracted, the dam body seepage prevention weak section, the pipeline welding seam concentrated section, the deep foundation pit excavation slope and the stress area of the pump station equipment foundation are marked as initial risk areas;

[0020] Based on the construction scheme risk annotation: read the risk level and control requirements marked in the construction scheme, and divide the high-altitude hoisting operation influence area, water and edge operation area, concrete curing critical period area and anti-seepage membrane laying area into initial risk areas;

[0021] Based on historical hidden danger records: retrieve historical inspection hidden danger data of the same project or similar projects, compare the location, type and rectification of hidden dangers, and divide the areas with leakage, structural settlement or welding defects and the areas that need to be continuously monitored after hidden danger rectification into key risk areas.

[0022] As preferred, the special detection of the risk area by the detection equipment carried includes:

[0023] When there is only an initial risk area, the dam body anti-seepage weak section and deep foundation pit excavation slope are divided into a first-level risk sub-area in the initial risk area, and the pipeline welding concentrated section and pump station equipment foundation stress area are divided into a second-level risk sub-area in the initial risk area, and the first-level risk sub-area is preferentially inspected;

[0024] When there are both initial risk areas and key risk areas, the key risk areas are preferentially inspected, and then the initial risk areas are covered in the order of first-level risk sub-areas and second-level risk sub-areas.

[0025] As preferred, the image data is preliminarily screened through the ground terminal, including:

[0026] Clarity screening: through the Canny edge detection algorithm based on gradient operator of the unmanned aerial vehicle ground control terminal, the edge features of the key parts in the image are extracted; the ratio of the total pixels of the key parts to the effective edge pixels is calculated, and when the ratio is lower than the preset clarity judgment threshold of the project, it is defined that the details of the key part structure cannot be identified by edge features, and the image clarity is unqualified;

[0027] Integrity screening: through the image stitching technology based on SIFT feature point matching of the ground control terminal, multiple images of the same inspection area are spliced into a regional panorama; the preset risk area GIS vector layer is called, and the overlapping area ratio of the panorama and the vector layer is calculated through spatial overlay analysis, and when the area of the uncovered area accounts for more than the preset integrity threshold of the project, the area of the vector layer, it is determined that the image integrity of the area is unqualified;

[0028] Key information correlation screening: through the image metadata analysis algorithm of the ground control terminal, metadata in the image is extracted, the metadata including a time stamp and GPS coordinates; if the analysis result shows that the time stamp field is missing, the GPS coordinate field is empty or the coordinate data exceeds the geographical range of the construction area, it is determined that the key information correlation of the image is unqualified.

[0029] As preferred, the re-shooting of the corresponding unqualified area includes:

[0030] For the area with unqualified definition: based on the topographic elevation data of the construction area, the flight height of the unmanned aerial vehicle is adjusted to the preset effective imaging interval of the project, and after synchronously calibrating the gimbal pitch angle or adjusting the sensitivity parameters of the imaging device, the unmanned aerial vehicle is controlled to perform re-shooting on the target area;

[0031] For the area with unqualified integrity: a local encryption flight path with the un-covered area as the core is generated through the ground control terminal, and the unmanned aerial vehicle is controlled to complete the re-shooting along the encryption flight path;

[0032] For the area with unqualified key information correlation: the flight log of the unmanned aerial vehicle is called through the ground control terminal, and the image file name is associated with the flight task identification, flight time stamp and GPS coordinates of the corresponding position in the flight log for supplementary recording; if there is no corresponding data in the flight log, a re-shooting instruction is generated and the unmanned aerial vehicle is controlled to return to the original collection area to re-collect images with complete metadata.

[0033] As preferred, the re-shooting of the corresponding unqualified area further includes:

[0034] After the re-shot images are transmitted to the ground control terminal, definition screening, integrity screening and key information correlation screening are performed on the re-shot images until all images meet the following conditions: the edge pixel ratio is not less than the definition determination threshold, the un-covered area ratio is not higher than the integrity threshold, and the metadata field is complete.

[0035] As preferred, the construction-related drawings include at least one of the water conservancy engineering construction blueprint, the three-dimensional BIM model or the GIS topographic data; and the construction key positions include at least dam structure, pipeline interface and pipeline weld.

[0036] As preferred, in the data collection step, when encountering adverse conditions, the unmanned aerial vehicle stops performing the task and returns to the departure place, the adverse conditions including extreme weather, device abnormality or strong interference signal.

[0037] As preferred, the format of the inspection data packet includes: image data and flight task identification arranged in the collection order.

[0038] In a second aspect, the application discloses an unmanned aerial vehicle-based intelligent inspection system for water conservancy project construction, which applies the unmanned aerial vehicle-based intelligent inspection method for water conservancy project construction.

[0039] The route planning module is configured to import construction-related drawings and analyze key construction positions, determine risk areas based on construction drawing analysis, construction scheme risk marking or historical hidden danger records, generate an inspection route for the key construction positions and the risk areas, dynamically adjust the coverage range of the inspection route according to the construction progress, and upload the generated inspection route to the flight control terminal of the unmanned aerial vehicle after the inspection route is audited and confirmed.

[0040] The unmanned aerial vehicle body comprises a power supply device, flight components, sensing components, a gimbal, a positioning module, a flight control terminal, imaging equipment, detection equipment and a flight control unit; the power supply device, flight components and sensing components are configured to cooperate to perform the inspection task, the positioning module is configured to meet the positioning requirements of the inspection after calibration, the flight control terminal is configured to receive the inspection route uploaded by the route planning module, the imaging equipment is configured to perform full-coverage imaging collection on the key construction positions and record the image data of each key construction position, the detection equipment is configured to perform special detection on the risk areas, and the flight control unit is configured to control the unmanned aerial vehicle to start the full-automatic inspection mode and fly according to the inspection route, and control the unmanned aerial vehicle to automatically suspend flight and perform a safe return procedure when encountering adverse conditions.

[0041] The AI identification system is configured to receive qualified images transmitted by the ground terminal, identify defects based on a preset defect library and mark the positions of the defects, classify the defects according to the severity of the defects, and generate an inspection log based on the identification results.

[0042] The ground terminal is configured to preliminarily screen the image data after the flight task is completed, initiate a re-shooting for unqualified areas, receive the inspection log generated by the AI identification system, integrate the image data, the inspection log and associated positioning data to generate an inspection data package, and upload the encrypted inspection data package to the supervision platform.

[0043] The supervision platform is configured to receive the inspection data package uploaded by the ground terminal, archive the inspection data package according to construction nodes, push serious defects to the construction unit and the person in charge of supervision, record the response time and processing progress of the defects, generate a defect disposal closed-loop tracking sheet, and keep the electronic log generated in each inspection for no less than a specified period.

[0044] Compared with the prior art, the unmanned aerial vehicle-based water conservancy project construction intelligent inspection method and system can effectively solve the problems of incomplete coverage, inaccurate identification, delayed response and data island in traditional inspection during the water conservancy project construction period, and has obvious optimization in inspection coverage, defect identification, data management, construction adaptation and safety benefits, and has the following beneficial effects.

[0045] The unmanned aerial vehicle performs full-automatic inspection mode to perform full-coverage imaging collection and special detection on key parts and risk areas of the construction, and realizes full-range dead-angle-free inspection of the construction area; the AI identification system automatically identifies defects, labels defect positions and classifies them according to severity, and generates an inspection log, thereby reducing the dependence on human experience, greatly improving the efficiency and accuracy of defect identification, and avoiding missed inspection and misjudgment; the ground terminal generates an inspection data packet and encrypts it to upload to the supervision platform, the supervision platform further archives the data according to the construction nodes, and correspondingly generates a closed-loop tracking sheet, thereby breaking the information island; and the flight path planning can dynamically adjust the coverage range according to the construction-related drawings and the construction progress, can accurately adapt to the dynamic construction rhythm, ensures the supervision of key construction procedures without blind spots, reduces the human investment and safety risks of manual inspection, and improves the accuracy, timeliness and traceability of construction supervision. Therefore, the technical scheme of the present application can fully meet the supervision needs of visual, quantifiable and traceable supervision during the water conservancy project construction period, improve the comprehensive benefits of construction supervision, and provide reliable protection for construction quality and safety. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical schemes in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 The flowchart of the unmanned aerial vehicle-based water conservancy project construction intelligent inspection method provided by the embodiments of the present application. DETAILED DESCRIPTION

[0048] The technical schemes in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0049] In this document, the terms "comprise" and "comprising" are used in the sense of "including" and alike, and allow for elements "not including" others that are not specifically listed.

[0050] The embodiment provides a method for unmanned aerial vehicle-based intelligent inspection of water conservancy construction in a first aspect. Figure 1 The method aims to solve the problems of limited coverage, low efficiency, scattered data and poor defect treatment closed loop of traditional manual inspection. The method comprises a flight path planning step, an inspection preparation step, a data collection step, a data identification step, a data uploading step and a closed loop management step.

[0051] In detail

[0052] The flight path planning step specifically comprises the following steps: analyzing construction key positions based on imported construction-related drawings, determining risk areas based on construction-related drawing analysis, construction scheme risk annotation or historical hidden danger records, generating inspection flight paths for construction key positions and risk areas, and dynamically adjusting the coverage range of the inspection flight paths according to the construction progress, and the generated inspection flight paths are uploaded to the flight control terminal of the unmanned aerial vehicle after being audited and confirmed.

[0053] Feasibly, the construction-related drawings comprise at least one of a water conservancy construction blueprint, a three-dimensional BIM model or GIS terrain data; and the construction key positions comprise at least dam body structure, pipeline interface and pipeline weld.

[0054] In this step, the determination of the risk area comprises at least one of the following:

[0055] Based on construction-related drawing analysis: extracting geological condition data, structure stress weak points and key process nodes in the construction-related drawings, and marking dam body seepage prevention weak sections, pipeline weld concentration sections, deep foundation pit excavation slope and pump station equipment foundation stress areas as initial risk areas;

[0056] Based on construction scheme risk annotation: reading the risk level and control requirements marked in the construction scheme, and marking high-altitude hoisting operation influence areas, water and edge operation areas, concrete curing critical period areas and impermeable membrane laying areas as initial risk areas;

[0057] Based on historical hidden danger records: calling historical inspection hidden danger data of the same project or similar projects, comparing the location, type and rectification of hidden dangers, marking areas with leakage, structure settlement or welding defects and areas that need to be continuously monitored after hidden danger rectification as key risk areas.

[0058] Further, the detection equipment carried is used to detect the risk area, including:

[0059] When there is only an initial risk area, the dam body seepage weak section and the deep foundation pit excavation slope are divided into a first-level risk sub-area in the initial risk area, and the pipeline welding concentrated section and the pump station equipment foundation stress area are divided into a second-level risk sub-area in the initial risk area, and the first-level risk sub-area is preferentially inspected; for the first-level risk sub-area: the dam body seepage weak section uses a thermal imager + infrared leak detector, the detection frequency is ≥3 times per construction day (additional 1 time after rainfall), and the temperature difference abnormal point and the seepage trace are focused on; the deep foundation pit excavation slope uses a high-definition imaging device + laser ranging module, the detection frequency is ≥2 times per excavation cycle, and the slope displacement and the supporting structure integrity are focused on; for the second-level risk sub-area: the pipeline welding concentrated section uses a macro imaging module + welding defect identification lens, the detection frequency is ≥2 times per 100 meters of welding completion, and the welding flatness, incomplete welding and welding tumor are focused on; the pump station equipment foundation stress area uses a high-precision imaging device, the detection frequency is ≥1 time per construction section of pouring completion, and the foundation surface crack, exposed steel bar and pouring density are focused on.

[0060] When there are both the initial risk area and the key risk area, the key risk area is preferentially inspected, and then the initial risk area is covered in the order of the first-level risk sub-area and the second-level risk sub-area; wherein the inspection of the key risk area includes: using corresponding equipment to detect: the seepage history area uses a thermal imaging + infrared detector, ≥3 times per day (additional 1 time after rainfall), to detect temperature difference abnormalities and seepage; the settlement history area uses a laser ranging + high-precision imaging device, the re-measurement interval is ≤4 hours, to detect height difference changes; the welding defect history area uses a macro imaging + welding module, ≥2 times per day, to detect welding flatness and secondary defects.

[0061] In a specific application, the implementation process of the route planning step of the embodiment can be: first, import the water conservancy construction blueprint, the three-dimensional BIM model or the GIS terrain data, identify the dam body structure, the pipeline interface, the pipeline welding and other key construction parts based on these data; then, determine the risk area in combination with the risk level (such as the high-altitude hoisting operation influence area) marked in the construction scheme or the seepage, welding defect area recorded in the historical inspection; subsequently, use DJIPilot, UgCS or Pix4Dcapture and other route planning software to automatically generate an inspection route for the dam body structure, the welding, the pipeline interface and other key areas, and dynamically adjust the route coverage range according to the construction progress; the generated route needs to be audited and confirmed by the project technical person in charge, and then uploaded to the flight control terminal of the unmanned aerial vehicle.

[0062] Based on the above route planning step design, by accurately identifying the key construction parts and risk areas based on construction drawings, risk labeling and historical records, the inspection route can accurately cover the core key parts such as dam structure, pipeline interface and various risk areas, avoiding the subjectivity and omission of traditional manual route planning, reducing invalid inspection range, and improving the pertinence of inspection; By generating targeted inspection routes and dynamically adapting to the construction progress, the route can be updated in real time with the construction progress (such as process advancement, area handover), ensuring that key areas in different construction stages can be effectively covered; At the same time, after being reviewed and confirmed, it is uploaded to ensure the scientificity and safety of the route, providing a reliable path basis for subsequent unmanned aerial vehicle full-automatic inspection, and reducing the risk of low inspection efficiency or data loss caused by unreasonable route from the source.

[0063] The inspection preparation step specifically includes: checking the working status of the power supply equipment, flight parts, sensing parts and gimbal of the unmanned aerial vehicle, and calibrating the positioning module of the unmanned aerial vehicle.

[0064] In a specific application, the specific configuration of the unmanned aerial vehicle of the embodiment can be: the power supply equipment uses ≥90% environmentally friendly lithium battery, equipped with 2 groups of backup batteries; the flight parts contain undamaged blades, the sensing parts contain clean attitude sensors, and the gimbal uses a three-axis gimbal with a stability of ±0.01°; the positioning module is an IMU+compass+RTK combined module with centimeter-level positioning; the flight control terminal can receive the encrypted route uploaded by the route planning module; the imaging device is a 20 million pixel RGB high-definition camera, and the detection device is a thermal imager with a resolution of 640x512; the flight control unit is built-in with an emergency return program, which can automatically abort the task and return when the weather is extreme or the equipment is abnormal.

[0065] Further, the implementation process of the inspection preparation step of the embodiment can be: when checking the power supply equipment of the unmanned aerial vehicle, ensure that the battery capacity is ≥90% and the number of backup batteries is ≥2 groups; when checking the flight parts and sensing parts, confirm that the blades are undamaged, the sensors are clean, and the gimbal is stable; when calibrating the positioning module, calibrate the IMU (inertial measurement unit), compass, and RTK module to ensure centimeter-level positioning accuracy; At the same time, insert an SD card with a remaining space ≥20GB, or connect an image back module.

[0066] Secondly, when the unmanned aerial vehicle is checked, the flight personnel preparation work is also completed, which specifically includes that the main operator needs to have a "unmanned aerial vehicle over-the-horizon driving license", and the auxiliary hand should be familiar with the terrain of the construction area, secondly, set up warning lines and warning signs (such as "unmanned aerial vehicle inspection, no entry" warning signs) in the operation area, clean up the construction personnel and high-altitude obstacles such as scaffolding under the inspection route, and finally check whether the intercom signal is clear and the emergency stop remote controller has sufficient power, to ensure that the communication and emergency control equipment are normally available.

[0067] Based on the design of the above inspection preparation step, by checking the state of the unmanned aerial vehicle power supply, flight, sensing and gimbal equipment, potential equipment faults (such as insufficient battery power, damaged propeller, abnormal sensor) are checked in advance, reducing the return of the unmanned aerial vehicle in the middle of the inspection process or the interruption of data collection due to equipment problems, and ensuring the continuity of the inspection task; by calibrating the positioning module, the unmanned aerial vehicle can obtain centimeter-level positioning accuracy, so that the collected image and geographic coordinates are accurately associated, providing a reliable spatial reference for subsequent defect positioning and area division; thereby improving the stability and data reliability of the unmanned aerial vehicle inspection, avoiding image blur, positioning deviation and other problems caused by poor equipment state, and reducing the cost of secondary inspection.

[0068] The data collection step specifically includes: the unmanned aerial vehicle starts the full-automatic inspection mode and follows the inspection route, collects images of key construction parts through the imaging equipment carried, and detects risk areas through the detection equipment carried; after the flight task is suspended, the ground terminal performs preliminary screening on the collected image data, controls the unmanned aerial vehicle to re-shoot the unqualified area, and after the re-shooting is completed, the unmanned aerial vehicle continues to perform the remaining inspection task or ends the flight task.

[0069] In this step, the image data is preliminarily screened by the ground terminal, including:

[0070] Clarity screening: through the Canny edge detection algorithm based on gradient operator of the unmanned aerial vehicle ground control terminal, the edge features of the key construction parts in the image are extracted; the ratio of the extracted effective edge pixels to the total pixels of the key parts is calculated, and when the ratio is lower than the preset clarity judgment threshold of the project, it is defined that the details of the key part structure cannot be identified by the edge features, and it is judged that the image clarity is unqualified;

[0071] Integrity screening: through the image stitching technology based on SIFT feature point matching of the ground control terminal, multiple images of the same inspection area are stitched into a regional panorama; the preset risk area GIS vector layer is called, and the overlapping area ratio of the panorama and the vector layer is calculated through spatial overlay analysis, and when the area of the uncovered area accounts for more than the preset integrity threshold of the project, it is judged that the image integrity of the area is unqualified;

[0072] Key information correlation screening: through the image metadata analysis algorithm of the ground control terminal, the metadata in the image is extracted, including the collection timestamp and GPS coordinates; if the analysis result shows that the timestamp field is missing, the GPS coordinate field is empty or the coordinate data is out of the geographical range of the construction area, it is judged that the key information correlation of the image is unqualified.

[0073] Further, the corresponding unqualified area is re-shot, including:

[0074] For the unqualified area of definition: based on the terrain elevation data of the construction area, the flight height of the unmanned aerial vehicle is adjusted to the preset effective imaging interval of the project, and after synchronously calibrating the gimbal pitch angle or adjusting the sensitivity of the imaging device, the unmanned aerial vehicle is controlled to perform re-shooting on the target area;

[0075] For the unqualified area of integrity: a local encryption flight path with the uncoated area as the core is generated through the ground control terminal, and the unmanned aerial vehicle is controlled to complete the re-shooting along the encryption flight path;

[0076] For the unqualified area of key information relevance: the flight log of the unmanned aerial vehicle is called through the ground control terminal, and the image file name is associated with the flight task identification, flight timestamp and corresponding GPS coordinates of the location in the flight log. If there is no corresponding data in the flight log, a re-shooting instruction is generated and the unmanned aerial vehicle is controlled to return to the original collection area to re-collect images with complete metadata.

[0077] Feasibly, the re-shooting of the corresponding unqualified area also includes:

[0078] After the re-shot images are transmitted to the ground control terminal, the re-shot images are subjected to definition screening, integrity screening and key information relevance screening until all images meet the following conditions: the edge pixel ratio is not less than the definition threshold, the uncoated area ratio is not higher than the integrity threshold, and the metadata field is complete.

[0079] Secondly, when encountering adverse situations, the unmanned aerial vehicle suspends the execution of the task and returns to the departure place, and the adverse situations include extreme weather, device abnormalities or strong interference signals.

[0080] Based on the design of the above data collection steps, through the full-automatic inspection of the unmanned aerial vehicle, the manual operation intensity is greatly reduced, and the inspection efficiency is improved by 5-10 times compared with the traditional manual step inspection. It is especially suitable for large-scale water conservancy projects such as long-distance water pipelines and large dams; through full-coverage imaging and special detection, the data comprehensiveness is ensured, covering the overall structure and focusing on the details of high-risk points; at the same time, combined with the ground terminal preliminary screening and re-shooting mechanism, real-time filtering of unqualified data is realized, and through parameter adjustment, encryption flight path re-shooting and other ways, the data quality entering the subsequent link is guaranteed, and the AI recognition misjudgment or omission caused by data defects is avoided.

[0081] The data recognition step specifically includes: after the flight task is completed, the qualified images are imported into the AI recognition system, the AI recognition system recognizes defects based on the preset defect library and labels the defect positions, and classifies them according to the defect severity, and generates an inspection log based on the recognition results.

[0082] In one embodiment, the specific configuration of the AI recognition system of the present embodiment can be: the system is deployed on an edge terminal or a supervision platform server, and is built-in with a YOLOv8 target detection model and a SAM image segmentation model; the preset defect library contains typical defect types of water conservancy construction period such as cracking, exposed steel bars, settlement, incomplete welding, structural misplacement, and poor welding; it can automatically label the defect GPS coordinates, and classify them into mild (not affecting the safety of the structure), moderate (requiring limited period of rectification), and serious (immediately stop work and rectify); and automatically generate an electronic inspection log containing the flight personnel, task number, flight route, and defect statistics.

[0083] Based on the design of the above data recognition step, the AI system recognizes defects and classifies them based on the preset defect library, replacing the traditional manual visual recognition. The deep learning model (such as YOLOv8) is used to quickly process massive images, and the defect recognition efficiency is improved by more than 80%, and the recognition accuracy (such as cracks, leakage, and welding defects) can reach more than 90%, reducing the subjective error of manual work. At the same time, the classification according to the severity makes the defect treatment priority clear, providing quantitative basis for subsequent pushing and rectification, and avoiding resource waste caused by different defects without differential treatment. Secondly, the standardized inspection log is automatically generated, integrating defect location, type, level and other information, realizing the structured output of the recognition result, and facilitating subsequent data integration and platform management.

[0084] The data uploading step specifically includes: the ground terminal integrates the image data, the inspection log and the associated positioning data to generate an inspection data package, and uploads the encrypted inspection data package to the supervision platform. The format of the inspection data package includes: image data arranged in the order of collection (each image is associated with the corresponding collection time, positioning data and defect label) and flight task identification.

[0085] In one embodiment, the specific configuration of the ground terminal of the present embodiment can be: the ground terminal is an industrial tablet loaded with a special inspection software, and is built-in with a Canny edge detection algorithm (for clarity screening), a SIFT feature point matching algorithm (for integrity screening), and a metadata analysis algorithm (for key information screening); it can receive the image data transmitted by the unmanned aerial vehicle through the wireless module and initiate a re-shooting instruction; after receiving the inspection log of the AI recognition system, it integrates the data in the format of “image + aerial photography time + GPS coordinates + recognition label + flight ID”, processes the data package through the AES-256 encryption algorithm, and uploads it to the supervision platform through the 4G / 5G network or a special interface.

[0086] Based on the design of the above data uploading step, the encrypted data packet is generated by integrating image, log and positioning data and uploaded to the supervision platform, ensuring the consistency of image, defect information and geographic coordinates, forming a complete chain of data, location and conclusion, avoiding information fragmentation; at the same time, the encryption processing (such as AES encryption) ensures the security of data in the transmission process, prevents the leakage or tampering of inspection data (including engineering sensitive information); in addition, the standardized data packet format (such as according to the collection order, containing task identification) facilitates the supervision platform to quickly analyze, archive, and provide structured data support for subsequent closed-loop management, improving data flow efficiency.

[0087] The closed-loop management step specifically includes: the supervision platform archives the inspection data packet according to the construction node, pushes the serious defects to the construction unit and the supervisor, records the response time and processing progress of the defects, and generates a defect disposal closed-loop tracking sheet; wherein the electronic log generated by each inspection is retained for no less than a specified period of time.

[0088] In one embodiment, the specific configuration of the supervision platform of the present embodiment can be: the platform is a web-based enterprise management system, which can receive encrypted inspection data packets, automatically archive data according to construction nodes (such as pipeline laying period, dam pouring period), provide query functions according to time, area and problem type; built-in message push module, real-time push serious defects to the system account and mobile APP of the project manager of the construction unit and the person in charge of the supervision party; automatically record the response time and processing progress of the defects, and generate a disposal closed-loop tracking sheet with a unique number; the electronic log retention time is set to 12 months, supporting PDF format download and export.

[0089] Based on the design of the above closed-loop management step, the inspection data is accurately matched with the engineering progress according to the construction node, forming a complete construction period quality archive, which has the function of backtracking query, meeting the engineering acceptance and accountability requirements; at the same time, the serious defect is pushed to the construction unit and the supervision party, ensuring the quick response of key problems, combined with the response time and progress record, forcing the responsible party to timely rectify and shorten the defect disposal period; in addition, the long-term retention of the closed-loop tracking sheet and the log realizes the traceability of the whole process of discovery, push, rectification and verification, forms a quality management closed loop, reduces the engineering safety risk caused by the defects not being handled in time, and at the same time meets the compliance requirements of industry supervision.

[0090] The present embodiment provides, in a second aspect, an unmanned aerial vehicle-based intelligent inspection system for hydraulic engineering construction, which applies the unmanned aerial vehicle-based intelligent inspection method for hydraulic engineering construction as described above, and the system comprises a flight path planning module, an unmanned aerial vehicle body, a ground terminal, an AI identification system and a supervision platform,

[0091] Specifically

[0092] The route planning module is configured to import construction-related drawings and analyze key construction sites, determine risk areas based on construction drawing analysis, construction scheme risk marking, or historical hidden danger records, generate inspection routes for the key construction sites and the risk areas, dynamically adjust the coverage range of the inspection routes according to the construction progress, and upload the generated inspection routes to the flight control terminal of the UAV after the inspection routes are verified and confirmed.

[0093] The UAV body includes a power supply device, flight components, sensing components, a gimbal, a positioning module, a flight control terminal, an imaging device, a detection device, and a flight control unit. The power supply device, flight components, sensing components, and gimbal are configured to cooperate to perform the inspection task. The positioning module is configured to meet the positioning requirements of the inspection after calibration. The flight control terminal is configured to receive the inspection routes uploaded by the route planning module. The imaging device is configured to perform full-coverage imaging collection on the key construction sites and record image data of each key construction site. The detection device is configured to perform special detection on the risk areas. The flight control unit is configured to control the UAV to start the full-automatic inspection mode and fly according to the inspection routes. When encountering adverse situations, the flight control unit controls the UAV to automatically suspend flight and perform a safe return procedure.

[0094] The AI recognition system is configured to receive qualified images transmitted by the ground terminal, identify defects based on a preset defect library and mark the positions of the defects, classify the defects according to the severity of the defects, and generate an inspection log based on the identification results.

[0095] The ground terminal is configured to preliminarily screen the image data after the flight task is completed, initiate a re-shooting for unqualified areas, receive the inspection log generated by the AI recognition system, integrate the image data, the inspection log, and associated positioning data to generate an inspection data package, and upload the encrypted inspection data package to the supervision platform.

[0096] The supervision platform is configured to receive the inspection data package uploaded by the ground terminal, archive the inspection data package according to construction nodes, push serious defects to the construction unit and the responsible persons of the supervision party, record the response time and processing progress of the defects, generate a defect disposal closed-loop tracking sheet, and retain the electronic log generated in each inspection for no less than a specified period.

[0097] It should be noted that the intelligent inspection system for water conservancy construction based on the UAV in the present embodiment corresponds to the intelligent inspection method for water conservancy construction based on the UAV described above. Therefore, the parts (including but not limited to specific implementation means, technical effects, etc.) not described in detail in the intelligent inspection system for water conservancy construction based on the UAV in the present embodiment can be correspondingly referred to the related description in the intelligent inspection method for water conservancy construction based on the UAV described above, which will not be described herein.

[0098] In the embodiments provided by the present application, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following: application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof. For software implementation, the procedures described herein can be implemented with associated hardware to perform the procedures described herein. In implementation, the procedures described above can be stored in a computer readable storage medium or transmitted as one or more instructions or code on a computer readable storage medium. The computer readable storage medium includes a computer storage medium and a communication medium, and the communication medium includes any medium that facilitates transfer of a computer program from one place to another. The storage medium can be any available medium that can be accessed by a computer. The computer readable storage medium can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage medium, or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0099] Finally, it should be noted that the above-described embodiments are merely exemplary of the application and should not be used in a limiting sense. Rather, the description is intended to cover any alternatives, modifications, and equivalents, which can be made to the embodiments described herein, by those skilled in the art in the spirit and scope of the present application.

Claims

1. A method for intelligent inspection of water conservancy engineering construction based on unmanned aerial vehicles (UAVs), characterized in that, The method includes: Flight route planning steps: Analyze key construction parts based on imported construction-related drawings, identify risk areas based on construction-related drawings, risk labeling of construction plans or historical hazard records, generate inspection routes for key construction parts and risk areas, dynamically adjust the coverage of inspection routes according to construction progress, and upload the generated inspection routes to the UAV's flight control terminal after review and confirmation. Inspection preparation steps: Check the working status of the drone's power supply equipment, flight components, sensor components and gimbal, and calibrate the drone's positioning module; Data acquisition steps: The drone starts the fully automatic inspection mode and follows the inspection route to collect full-coverage images of key construction parts through the onboard imaging equipment, and conducts special inspections of risk areas through the onboard detection equipment; after the flight mission is suspended, the ground terminal performs a preliminary screening of the collected image data, and controls the drone to re-shoot unqualified areas. After the re-shoot is completed, the drone continues to perform the remaining inspection mission or ends the flight mission. Data recognition steps: After the flight mission is completed, qualified images are imported into the AI ​​recognition system. The AI ​​recognition system identifies defects and marks the defect locations based on a preset defect library. At the same time, it classifies the defects according to their severity and generates an inspection log based on the recognition results. Data upload steps: The ground terminal integrates the image data, inspection logs and associated positioning data to generate an inspection data package, and then encrypts the inspection data package before uploading it to the monitoring platform; Closed-loop management steps: The monitoring platform archives the inspection data package according to the construction nodes, pushes serious defects to the responsible persons of the construction unit and the supervision party, records the response time and processing progress of the defects, and generates a closed-loop tracking form for defect handling; the electronic log generated for each inspection shall be retained for no less than the prescribed period.

2. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The determination of the risk area includes at least one of the following: Based on the analysis of construction-related drawings: extract geological condition data, structural weak points and key process nodes from the construction-related drawings, and delineate the weak sections of the dam body seepage prevention, the concentrated sections of pipeline welds, the deep foundation pit excavation slopes and the stress areas of the pump station equipment foundations as initial risk areas; Based on the risk labeling of the construction plan: read the risk level and control requirements clearly marked in the construction plan, and delineate the high-altitude hoisting operation impact area, water and edge operation area, critical concrete curing area and geomembrane laying area as initial risk areas; Based on historical hazard records: retrieve historical inspection hazard data for the same or similar projects, compare the location, type, and rectification status of the hazards, and designate areas where leakage, structural settlement, or welding defects have occurred, as well as areas that require continuous monitoring after hazard rectification, as key risk areas.

3. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 2, characterized in that, The aforementioned specialized detection of risk areas using onboard detection equipment includes: When only the initial risk area exists, the weak section of the dam body and the deep foundation pit excavation slope are classified as the first-level risk sub-area in the initial risk area, and the concentrated section of pipeline welds and the stress area of ​​the pump station equipment foundation are classified as the second-level risk sub-area in the initial risk area. The first-level risk sub-area is inspected first. When both initial risk areas and key risk areas exist simultaneously, priority should be given to inspecting the key risk areas, and then the initial risk areas should be covered in the order of first-level risk sub-areas and second-level risk sub-areas.

4. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The aforementioned preliminary screening of image data via a ground terminal includes: Sharpness screening: The edge features of key construction parts in the image are extracted using the Canny edge detection algorithm based on gradient operators on the UAV ground control terminal; the ratio of the extracted effective edge pixels to the total pixels of the key parts is calculated. When this ratio is lower than the preset sharpness judgment threshold of the project, the details of the key part structure cannot be identified by edge features, and the image is judged to be unqualified in sharpness. Integrity screening: Using SIFT feature point matching-based image stitching technology on the ground control terminal, multiple frames of images of the same inspection area are stitched together to form a panoramic image of the area; the preset risk area GIS vector layer is called, and the overlap area ratio between the panoramic image and the vector layer is calculated through spatial overlay analysis. When the proportion of the uncovered area to the total area of ​​the vector layer exceeds the preset integrity threshold of the project, the image integrity of the area is determined to be unqualified. Key information correlation screening: Metadata in the image is extracted through the image metadata parsing algorithm of the ground control terminal. The metadata includes the acquisition time stamp and GPS coordinates. If the parsing result shows that the time stamp field is missing, the GPS coordinate field is empty, or the coordinate data exceeds the geographical range of the construction area, the key information correlation of the image is determined to be unqualified.

5. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 4, characterized in that, The aforementioned reshooting of the corresponding defective areas includes: For areas with unsatisfactory clarity: Based on the terrain elevation data of the construction area, adjust the drone's flight altitude to the preset effective imaging range of the project, simultaneously calibrate the gimbal pitch angle or adjust the sensitivity parameters of the imaging equipment, and then control the drone to retake the target area. For areas with incomplete coverage: a localized encrypted flight path is generated with the uncovered area as the core through the ground control terminal, and the drone is controlled to complete the supplementary shooting along the encrypted flight path; For areas with unqualified key information correlation: retrieve the UAV flight log through the ground control terminal, associate the image file name with the flight mission identifier, flight timestamp and corresponding GPS coordinates in the flight log, and supplement the data; if there is no corresponding data in the flight log, generate a supplementary shooting command and control the UAV to return to the original acquisition area to re-acquire images with complete metadata.

6. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 4, characterized in that, The aforementioned reshooting of the corresponding defective areas also includes: After the re-captured images are transmitted to the ground control terminal, they undergo sharpness screening, integrity screening, and key information correlation screening until all images meet the following requirements: the proportion of edge pixels is not lower than the sharpness judgment threshold, the proportion of uncovered area is not higher than the integrity threshold, and the metadata fields are complete.

7. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The construction-related drawings include at least one of the following: water conservancy project construction blueprints, 3D BIM models, or GIS terrain data; the key construction components include at least the dam structure, pipe joints, and pipe welds.

8. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, In the data acquisition step, if an adverse situation is encountered, the UAV will suspend its mission and return to its starting point. The adverse situation includes extreme weather, equipment malfunction, or strong interference signals.

9. The intelligent inspection method for water conservancy engineering construction based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The format of the inspection data packet includes: image data arranged in the order of acquisition and flight mission identifier.

10. A UAV-based intelligent inspection system for water conservancy engineering construction, employing the UAV-based intelligent inspection method for water conservancy engineering construction as described in any one of claims 1-9, characterized in that, The system includes a flight path planning module, the drone itself, a ground terminal, an AI recognition system, and a monitoring platform; The flight path planning module is used to import construction-related drawings and analyze key construction parts. Based on the analysis of construction drawings, risk labeling of construction plans or historical hidden danger records, risk areas are identified. Inspection routes are generated for key construction parts and risk areas. The coverage of the inspection routes is dynamically adjusted according to the construction progress. The generated inspection routes are uploaded to the UAV's flight control terminal after being reviewed and confirmed. The UAV body includes a power supply, flight components, sensing components, a gimbal, a positioning module, a flight control terminal, an imaging device, a detection device, and a flight control unit. The power supply, flight components, sensing components, and gimbal are used to cooperate in performing inspection tasks. The positioning module is used to meet the inspection positioning requirements after calibration. The flight control terminal is used to receive the inspection route uploaded by the route planning module. The imaging device is used to perform full-coverage imaging of key construction areas and record image data of each key construction area. The detection device is used to perform special detection of risk areas. The flight control unit is used to control the UAV to start a fully automatic inspection mode and fly according to the inspection route. In case of adverse conditions, it controls the UAV to automatically stop flight and execute a safe return procedure. The AI ​​recognition system is used to receive qualified images transmitted by ground terminals, identify defects based on a preset defect database and mark the defect locations, classify them according to the severity of the defects, and generate inspection logs based on the recognition results. The ground terminal is used to perform initial screening of image data after the flight mission is completed, initiate reshoots for unqualified areas, receive inspection logs generated by the AI ​​recognition system, integrate image data, inspection logs and associated positioning data to generate inspection data packets, and encrypt the inspection data packets before uploading them to the monitoring platform. The monitoring platform is used to receive inspection data packets uploaded by ground terminals, archive the inspection data packets according to construction nodes, push serious defects to the responsible persons of the construction unit and the supervision party, record the response time and processing progress of defects, generate a closed-loop tracking form for defect handling, and retain the electronic log generated by each inspection for no less than the prescribed period.

Images